Epigenetic silencing involving alterations in DNA methylation and chromatin structure at promoter region CpG islands is a common mechanism of tumor suppressor gene inactivation in human cancers. However, the mechanisms underlying this event remain poorly understood. An unresolved question is why are some genes targets of aberrant methylation in human cancers while others are never affected? In preliminary work, we have shown that even in the context of an increased cellular capacity for de novo methylation, CpG islands differ in their potential for aberrant methylation. By applying DNA pattern recognition and machine learning techniques, we have developed an algorithm based on several short sequence patterns that is capable of accurately discriminating methylation-prone and methylation-resistant CpG islands. These studies indicate that the epigenetic status of a CpG island can be predicted based on DNA sequence features, and lead us to propose that one factor contributing to the non-random patterns of CpG island methylation observed in human tumors is an underlying susceptibility conferred by local sequence context. The goal of this proposal is to define the genomic signature associated with aberrant methylation. The long term objectives are (i) to identify and to functionally characterize local sequence attributes that contribute to the propensity towards, or protection from, aberrant methylation, and (ii) to develop and to test novel tumor specific classifiers capable of predicting genomic loci at risk of aberrant methylation. Specifically, we will determine whether sequence features identified in silico act in cis to promote or to prevent de novo methylation in vivo using an episomal transgene approach. In preliminary work, we have identified a relationship between methylation-prone CpG islands and genomic regions bound by the polycomb repressor complex. As a second component of the project, we will determine the role of PRC2 in methylation susceptibility. As a third component of this project we will refine our computational models by 1) determining whether CpG islands predicted to be methylation-prone are in fact targets of aberrant methylation in human cancers, 2) using this information to re-train the prediction model, and 3) developing and testing a novel lung cancer specific classifier based on large-scale CpG island methylation data from primary lung tumors. We anticipate that the information gained from these studies will allow for a better understanding of the mechanisms underlying the epigenetic silencing of tumor suppressor genes that accompanies carcinogenesis. Moreover, the ability to predict the methylation status of CpG islands genome-wide will provide an important resource for the identification of novel gene targets for further study as potential cancer biomarkers. PUBLIC HEALTH RELEVANCE: The overarching goal of this research is to better understand the mechanisms by which important growth regulatory genes are targeted for epigenetic silencing in human cancers. Although the consequences of this event are the same as a mutation, there is no molecular defect in the DNA sequence itself. Rather, there is a problem with methylation marks on the DNA and the histone proteins it is wrapped around that render certain genes abnormally and permanently silent. We plan to use a computational approach to identify DNA features that put certain genes at risk of aberrant methylation. Because genes that are silenced by methylation are otherwise structurally sound, the potential for reactivating these genes by blocking or reversing the methylation process represents an exciting molecular target for chemotherapeutic intervention. A better understanding of the factors that contribute to aberrant methylation, including the identification of sequence features that attract or repel DNA methylation, will be an important step in achieving this long-term goal. Moreover, the ability to identify those genes at risk of epigenetic silencing will provide novel molecular targets for further study as potential biomarkers for improved cancer diagnosis and treatment planning.